Temporally Consistent Tone Mapping of Images and Video Using Optimal K-means Clustering

Magnus Oskarsson

doi:10.1007/s10851-016-0677-1

Abstract

The field of high dynamic range imaging addresses the problem of capturing and displaying the large range of luminance levels found in the world, using devices with limited dynamic range. In this paper we present a novel tone mapping algorithm that is based on K-means clustering. Using dynamic programming we are able to not only solve the clustering problem efficiently, but also find the global optimum. Our algorithm runs in hbox {O}(N^2K) for an image with N input luminance levels and K output levels. We show that our algorithm gives comparable results to state-of-the-art tone mapping algorithms, but with the additional large benefit of a minimum of parameters. We show how to extend the method to handle video input. We test our algorithm on a number of standard high dynamic range images and video sequences and give qualitative and quantitative comparisons to a number of state-of-the-art tone mapping algorithms.

Highlights

The human visual system can handle massively different levels in input brightness
We address the problem of tone mapping as a clustering problem
If running times are a priority, we propose the use of key frames, where the tone mapping function is estimated for each key frame using the dynamic programming scheme, and for intermediate frames the tone mapping function is interpolated linearly between the two nearest key frames

Summary

Introduction

The human visual system can handle massively different levels in input brightness This is necessary to cope with the large range of luminance levels that appear around us—for us to be able to navigate and operate in dim night light as well as in bright sun light. The HDR field has grown, and today, many camera devices have built in functionality for acquiring HDR images. This can be done in hardware using sensors with pixels that can capture very large differences in dynamic range. Depending on the application the role of the tone mapping operator (TMO) can be different, but in most applications the ability to capture both detail in darker areas and very bright ones is important. Tone mapping can be an important component in image enhancement for, e.g., images taken under poor lighting [34,59]

Methods

Results

Conclusion